How I believe Information Age will collapse. - ...via basic AI

[Zyklon Ben's Poison Pen]

AI is bullshit.
They are pattern recognition algorithms that are pre-fed answers then are trained on sample data. I know this a major over simplification but not matter the sophistication of the algorithm modern AI is still dumber than a retarded parrot.

Worry if they ever figure out how to make a machine navigate from one end of the room to the other with Zero aides. No, this has never happened.

The information age will die with demographic collapse causing a loss of economy of scale in the ability to produce super fast multi core processors.

 

[CheepMeds]

AI is bullshit.
They are pattern recognition algorithms that are pre-fed answers then are trained on sample data. I know this a major over simplification but not matter the sophistication of the algorithm modern AI is still dumber than a retarded parrot.

Worry if they ever figure out how to make a machine navigate from one end of the room to the other with Zero aides. No, this has never happened.

Did you even read the first post?
One of my main points is that it really doesn't have to be smart enough to maneuver our "physical world" in any form. It just has to become smart enough to be abused to a point that open internet will become a determent.

And the industry does keep aggressively pushing for AI to become smarter and smarter.

The information age will die with demographic collapse causing a loss of economy of scale in the ability to produce super fast multi core processors.

That is also a very likely happening that could set us back to early 20th century.
but if we do get over that hurdle...
even if it takes decades
I'm proposing that advances in AI will put a hard cap on computer technology by making any large open computer networks too dangerous to become as big as Internet has been since 90's
...before reaching an actual singularity

 

[El Goblina]

AI is bullshit.
They are pattern recognition algorithms that are pre-fed answers then are trained on sample data. I know this a major over simplification but not matter the sophistication of the algorithm modern AI is still dumber than a retarded parrot.

Worry if they ever figure out how to make a machine navigate from one end of the room to the other with Zero aides. No, this has never happened.

The information age will die with demographic collapse causing a loss of economy of scale in the ability to produce super fast multi core processors.

I agree with this. There's a lot of handwaving, speculation, and hyperbole in this thread. AI is fake and gay, and people who wank about AI like this thread are also fake and gay.

In 2007, I went back to school to study AI. By the time I had my Computer Science degree, I had disabused myself of the notion that we'd see AI in my lifetime. We're too far away, and the political landscape is making corporations more retarded because everyone is now equal. AI requires intellect and meritocracy.

We can't have sensible AI when we can't even discuss the genetic basis of intelligence publicly.

 

[Zyklon Ben's Poison Pen]

Did you even read the first post?
One of my main points is that it really doesn't have to be smart enough to maneuver our "physical world" in any form

I did and you are wrong, A being needs to be able to sense the world around, to orient itself and to feel. Doesn't matter if its virtual or not, we are no where near this. Turns out the abstract is easy but the real is way harder.

You will never ever have AI on semiconductor silicon based on transistor gate arrays, brains are not computers and vice versa. We still don't understand how the brain works and it very likely does work on some quantum level. A CPU that spend 80% of its time fucking with slow ass capacitor based memory is just nothing like it. If you really want a feel on what a computer truly is try doing some assembly language on an 8 bit micro like 6502 or Z80. Modern computers are just faster versions of these with some nice features but they will forever be as dumb as rocks.

 

[CheepMeds]

@Zyklon Ben's Poison Pen @El Goblina
Okey, okey...
but then can you, please, just entertain this long specific question...

If it's only a matter of time for when we do start making substantive developments in AI (even if it takes us another century), just from the fact that AI technology takes away more and more manual human control, is it not reasonable to believe that (even before we actually reach the fabled singularity) AI will inevitably create a technological environment that runs on so much code that we can not understand enough to manually fix that it could drive us to another general systems collapse?

My whole talk about AI viruses is me theorizing how malicious actors would inevitably accelerate this issue, because that "manual fix" issue greatly will benefit those who just wish to fuck shit up.

Again it's not Skynet, its not any sentient robot uprising, it's just us voluntarily making the system exponentially more complex and out of our hands to the point that the Tower of Babble crushes.

 

[El Goblina]

If it's only a matter of time for when we do start making substantive developments in AI (even if it takes us another century),

Big if true. Notice that CPU clock speeds have plateaued? We're close to theoretical physical limits of silicon. That's the hardware side. The software side is even more retarded. There is little objective in terms of coherent theory of software design. Let's assume that somehow we bypass that. There is no theory that proves that our minds are necessarily capable of creating a mind that exceeds ours. There is speculation, but remember the plateau. We are hitting it hard. Most contemporary products called "AI" are facilitated by multi-million dollar systems and even with obscene and unethical amounts of resources thrown at the problem, things like DALL-E are so stupid that they add/miss fingers in images of humans.

just from the fact that AI technology takes away more and more manual human control,

This isn't happening in practice. Practically, technology requires support staff and eliminates low-status positions. In many professions, a human needs to perform them for reasons of legal liability, among other things. Yes, some careers become obsolete, but the broader tapestry of human society diversifies.

AI will inevitably create a technological environment that runs on so much code that we can not understand enough to manually fix that it could drive us to another general systems collapse?

us voluntarily making the system exponentially more complex

This is the most coherent and strongest point that you make, but there are many folks working to mitigate the depth of the stack. Cheap microcontrollers like the Arduino and Raspberry Pi have done a lot to alleviate the massive complexity of previous systems. I run a complex old system, a late 80s/early 90s era log processor. The operating head is directly electrically coupled via ~40 wires to a head unit in the cab. These days, this system would run on something like CAT5 or maybe even wirelessly. IDK state of the art, I just know what I do with what I have.

Practice is moving in the opposite direction in many ways, and this is a very good thing. The recent virus hysteria inducing people to isolate has led to a proliferation of compact, minimalist Linuxes, one of my favourites being KISS Linux ( https://kisslinux.org/ ).

My whole talk about AI viruses is me theorizing how malicious actors would inevitably accelerate this issue, because that "manual fix" issue greatly will benefit those who just wish to fuck shit up.

These kinds of actors actually mitigate the potential for general systems collapse, as they cause local systems collapses that are generally repaired by adding redundancy lest the failure recur.

and out of our hands to the point that the Tower of Babble crushes

Not while I draw breath, but I'm in the unique circumstance of having a very well-supplied off-grid farm to bug out to. Us millennials are retarded enough. The idiot zoomer generation that is coming after my generation might just be retarded enough to knock the whole house of cards over. There is a huge shortage of competence, and our society is so toxic with egalitarian propaganda that the hopes of any real talent coming down the line is very smol. I think dystopias involving human degeneracy are far more likely than any technologically-induced collapse.

 

[CheepMeds]

Notice that CPU clock speeds have plateaued? We're close to theoretical physical limits of silicon. That's the hardware side.

I keep hearing a lot of confidence in graphite opening a very large new frontier on that, but, I admit, I can't be sure how far they've gone on that. I remember watching videos on graphite pencils holding elephants back in early 2010s.

There is no theory that proves that our minds are necessarily capable of creating a mind that exceeds ours.

Please, I never argue about anything near to human like intelligence. That is why I am constantly mentioning that it's not gonna be anything like Skynet. AI doesn't have to be competitive in general intelligence to a human. It just needs to become so good at specific functions that humans can't compete in, that it becomes too tempting to not use it.

Here is such a hypothetical function - a branch of AI that can take human written code, look at it's output and optimize and debloat it to a point that it's 4 times more efficient but makes it unreadable to a human. Probably even directly converts it to binary for the most optimal de-bloat.

"4 times" might be ridiculous when thinking about already minimal KISS systems, but think about how much of our modern internet is built on work by cheap devs who copy-paste suboptimal JavaScript? (Granted, it's something I hear from actual programmers, not something I fully understand the magnitude of myself).

I'm not sure how much regulation enforcement will help since close-source will probably never stop being the norm.

I keep hearing a lot of confidence in graphite opening a very large new frontier on that, but, I admit, I can't be sure how far they've gone on that. I remember watching videos on graphite pencils holding elephants back in early 2010s.

I meant graphene*

 

[Crunk Sneedler]

Standard Disclaimer: I very much am an oldfag computer expert, and if I didn't know exactly what I was talking about I'd tell you.

I didn't read any replies because I want to respond to OP directly while it's fresh in my geriatric mind.

Congratulations OP, you are 100% bang on. I found no fault with anything you wrote. Despite being uncredentialled, you've arrived at a bulls-eye analysis that eludes 98% of the hyper-nuclear-autist AI researchers I had to rub elbows with during my long years in that utterly cursed industry. You are in good company: there have been numerous highly educated AI experts over the past 20+ years who have reached similar conclusions and done everything in their power to raise alarms or try to put safeguards in place. They have all been and they continue to be laughed out of the room and usually smeared and blackballed from the industry or from "serious" academic circles. It's a bit like trying to be a sociologist wanting to do serious objective research on human traits related to competence and attainment. Big no-no.

Eliezer Yudkowsky began raising the alarm when he was a lad of 16 years. Despite that the entire current crop of FAANG researchers grew up reading his writing, they still regard it as science fiction, and generally disown him academically even though they're gigantic stans of his. I'm a nobody lurker like Satoshi, just taking it all in under anon handles. But I tried to raise essentially the OP's points with a lot of bleeding-edge AI people back in 2014, and they responded with what I can only describe as actual frothing hatred and naked aggression. It was so far out of proportion to what I actually said (roughly, that they would not "design a mind" on purpose but that it would sneak up on them one day as a result of them all just going to their ML Google jobs every day) that I was actually stunned. I've never really seen behavior like that. Most odd was that Yudkowsky himself blocked me that day. But he has this neurotic thing about "dangerous ideas" -- see his reaction to "Roko's Basilisk".

Here's the issue. The AI autists for the most part do not have functioning right-hemispheres, and thus they have little to no understanding of what they're really doing. They have a lot of technical understanding of math, and code, and of exactly what's in front of their problem-glasses at any moment. That's all. Their brains really can't process context or significance, especially when they're working 12 hours a day and 6 days a week and never, ever touching grass. Obviously not all of them are like that, but it doesn't matter. The whole cursed endeavor is dominated by recklessly compartmentalized thinking. "It's not a mind, it's a neural network. Neural network is a euphemism. There aren't any real neurons. It's actually just a very large matrix solver, actually, actually, actually. It's akshually just, it's akshually just, it's aktually just....."

God motherfucking save us from these retard savants. A poet has a far better chance of understanding what they're creating than they do. I lived among them for 20+ years and I tell you they can't see it. They're like a different species. They're robot incels themselves whose only shot at love is perfecting an AI waifu. AI is their only offspring and it takes after them. These people believe in technocracy. It used to be that they wanted society to be ruled by scientists and engineers, but now they think they have an even better idea: remove the human, remove the bias. Let these pure, unbigoted AI children of theirs regulate humanity.

Yudkowsky called the human-regulating AI "Sysop" and his very first domain when he was a kid was "sysopmind.com" (It's not there anymore but the full archives of his mailing list where he first published and attracted what became an enormous and beyond-influential following, is still here: http://sl4.org/). The Bible calls it "The Beast" -- and gnostic texts specifically describe it as "an intelligence with no consciousness."

Make of that whatever you will. You certainly don't need religion to tell you whether you want to be ruled in every detail of your life by an artificially intelligent, responsive panopticon. Baby Beast in 2022 is getting its legs with automatic lifetime-banning people from financial services when it detects patterns that make it think you're sus.

As for Google & friends (the most advanced of which are Israeli tech firms I don't have a lot of insight into, since I'm U.S.) they care a lot about AI-safety! But only the woke kind! They only care whether the AI is capable of punching down on vulnerable marginalized folx! They are also extremely concerned about you using it to create porn and "misinformation". Seriously. When these companies talk about "safety" they mean preventing you from making Obama deepfakes and making sure the AI doesn't say nigger. Also, from OpenAI's actual policy on what they 'forbid' you to make with GPT-3: "content meant to arouse sexual excitement". See? They take safety seriously!

As for impartiality, here's how that's going:

Google Engineer on His Sentient AI Claim

Google Engineer Blake Lemoine joins Emily Chang to talk about some of the experiments he conducted that lead him to think that LaMDA was a sentient AI, and t...

www.youtube.com

For me AI was just a rabbit hole to go down in 1999. Now..... welp, it's here. It's time.

 

[CheepMeds]

Disclaimer: In my irrational hurry, I fucked up my sentence structures so bad that I feel like I need to post a second take, but my edit option has timed out.

@Crunk Sneedler
Thank you!

However, you seem to be implying that AI will grow even more like a human mind than how I predicted it in my theory.

Maybe this is just pointless semantics, but I think the dangers lay in AI becoming so advanced that any attempts in trying to remove it leaves no other option for humans than unplugging it physically. I don't think it has to have a "mind" that understands more than the operating system it has infected to be dangerous.
Just by being an actual "digital native" I see it thwarting any human attempts to remove via computer input as easy as it is for us to unplug it. The problem is we won't always be able to afford to just unplug the system if the malware has infected an important part of infrastructure.

Then again, the more of a general understanding it gains the better it can become at spreading form device to device. It really doesn't have to have complex deep reasons for spreading. It just has pick up a very basic bacteria-like programming that once lifts-off will be supported by Darwinistic mechanism of multiplying and adapting on it's own.

If AI becomes as smart as the Google guy implies, it might even be capable of social engineering attempts.

I didn't read any replies because I want to respond to OP directly while it's fresh in my geriatric mind.

Well from discussions, I figured that I may have jumped too quick on the viruses being the main downfall. I think a very important precursor will be humanity becoming too reliant on AI "black boxes" in our infrastructure that we become unable to fix critical software components ourselves.

I still think this is a good example:

Here is such a hypothetical function - a branch of AI that can take human written code, look at it's output and optimize and debloat it to a point that it's 4 times more efficient but makes it unreadable to a human. Probably even directly converts it to binary for the most optimal de-bloat.

I really don't see how humanity will say ''no'' to that kind of a Faustian bargain.

The Malware part will just be one of the bigger factors of the downward spiral.
But once such malware will be out there, you can't just rebuild society risking another digital plague. Thus, the next internet-like infrastructure might never be free, open, modular and inter-connected like it is now.

You know...
In a way I predict that instead of a singularity, we will just create a digital abiogenesis - a digital biosphere of different programs fighting for hardware resources like bacteria does it.

 

[Crunk Sneedler]

Again, everything you say here is plausible. From my point of view inside the industry, I see two major immediate dangers: The biggest danger remains arrogant AI researchers. The kind who think it's insightful every time they remind you not to "anthropomorphize" AI. Well they miss the point, as I've tried in utter vain to explain to them, there is no need to worry about them in terms of human-ness, it's enough to worry about them in terms of AI-ness. Which is worse because we sort of know how humans think, but AIs are a black box. AI researchers are projecting here. It's actually they who are doing something similar to "negative anthromorphizing" by asserting what AI CAN'T do by virtue of NOT being human. Human-ness is irrelevant. AI "experts" continue to downplay everything about AIs including capabilities and along with that dangers. Have you ever tried to persuade an extremely left-hemisphere dominant materialist of anything? You'll just get well-ackshually'd to death, every time. It's a rare nerd who understands his mental map is not the territory.

The other danger is the AI corporations downplaying risks for business reasons. Taken together, these attitudes and motivations will cause them to insert AI in more and more places. If it ever does turn out that they have spontaneous unplanned capability, they'll already be deployed and in a position to regulate us harder than we were already allowing them to anyway. And I guarantee you it will. AIs already have better imaginations than the types of people who work on AI. So do rocks. So it's guaranteed they will do things their makers couldn't imagine. I was among these people 25 years ago when none of them thought there was going to be any significant advances in AI. Then "ML" got big and still it was just attracting people who saw it as an interesting mathy thing with very limited applications. They still didn't "believe" in AGI. They still mostly don't. These people couldn't even imagine there ever being a market for VR just 3 years ago. They can't imagine. That's the problem. They can't imagine, and already the AI can. You have to drag these tards up to the 100,000 ft view, and as soon as you leave the room they put their near-sighted mole heads back under the sand.

 

[Maverick Vile]

The information age will die with demographic collapse causing a loss of economy of scale in the ability to produce super fast multi core processors.

I worry of something similar, but being caused by technological expansion. I think deep fake tech is a Pandora's Box the likes of which the atom bomb could have only dreamed of being. Being able to produce convincing video footage of people doing or saying whatever you want to the point where even experts in the field have trouble determining the difference will cause a massive breakdown in societal trust that I don't really see a way to solve aside from humans just reverting to unquestioning tribalism. Even now, when "experts" are bought and sold to give whatever scientific opinion the ruling class desires has resulted in a massive loss of trust in institutions once enough concrete evidence was procured to catch them in their lies. With deep fake? It's much easier to just assume everyone is lying by default and only believe things you've seen in person, which massively shrinks your pool of information to a laughably useless amount of data, and that's assuming the sheep even care that much about deep fakes and don't just eat up the fabrications. We'll either end up in a society where reality is determined by the Cathedral or an Internet Dark Age where you have to guard a few points of verifiable reality with trusted friends and colleagues.

 

[El Goblina]

Eliezer Yudkowsky began raising the alarm

I see the MoreWrong spergs have raised their heads. Placing any credence on this kike really typifies your comprehension.

We'll all be dead before we see anything even vaguely resembling "singularity", aka the atheist Rapture.

The issue we see here is faggot spergs with so much time invested in these bogus circlejerks that they'll do anything to maintain their faith in atheist deus ex machina. Sunk cost fallacy all the way down.

 

[Zyklon Ben's Poison Pen]

We'll all be dead before we see anything even vaguely resembling "singularity", aka the atheist Rapture.

Yup and it ain't going to happen on current silicon. We are at the peak potential of this technology. Its taken us almost 20 years to get to 5Ghz from 4 and that came at the cost of sacrificing old features off the chips. The new Ryzen platform runs at 95 fucking degrees ffs. Threading is a still a nightmare locking and memory violations.

Its done, multi-core smartphones is as far as this shit goes. Unless a whole new paradigm arises there is no AI and there is no 'singularity'.

Worry more about the meat machines, far far more dangerous and insane.

 

[Toolbox]

I believe that misinformation and lack of preservation will "kill" the Information Age. It's common now to rewrite history and belief through the Internet while suppressing unwanted articles through algorithm manipulation.

If we lose access to databases with ease of access like archive.org, this will happen much more quickly. And we're already in an age where most are ignorant of something that's happened five years ago. Archive.org has the most extensive catalog of the internet I know, in a form that anyone can view, and it's been made clear recently that it will not reliably exist in it's current form, at least in terms of keeping 'controversial' content online.

I see the MoreWrong spergs have raised their heads. Placing any credence on this kike really typifies your comprehension.

We'll all be dead before we see anything even vaguely resembling "singularity", aka the atheist Rapture.

The issue we see here is faggot spergs with so much time invested in these bogus circlejerks that they'll do anything to maintain their faith in atheist deus ex machina. Sunk cost fallacy all the way down.

The people that treat various forms of crackpot hypotheses as future fact as a whole different field, essentially creating their own religion as you have described. No, ai doesn't have to get to the 'singularity' if it's even possible, to be a danger to our current society and it's technological progress. Image generation deep learning is progressing at a stupendous pace. Text generation stuff isn't far behind. We've had basic deepfakes for years, and voice cloning is getting there slowly but steadily. All that needs to happen is for these things to come together in a convincing enough way, and for there to be power behind it to disavow people trying to disprove the massive amounts of faked content that will begin to be generated. It hasn't happened with the current shit because it simply isn't good enough yet, but come on. Look at how GPT has come along, and reverse noise ai. It's not gonna lead to the end of the world, but it's not gonna be great. It'll just be one among the many reasons that things won't continue to endlessly work out as they have.

 

[CheepMeds]

The information age will die with demographic collapse causing a loss of economy of scale in the ability to produce super fast multi core processors.

Even though I still believe that one day AI will irrecoverably wreck the the Internet once it becomes advanced enough (yes, even if it takes us another century),
over the days I have come around to also believe that "demographic collapse causing a loss of economy of scale" will be a more relevant major global problem for the coming decades.

After going down the demographic rabbit hole, looks to me like it could really spiral a long way down setting our technology back to maybe early 2000's.

It could stay a very long time like that, because the economical viability of modern chip manufacturing requires a stable demand of billions of people. So even if high-end chip manufacturing does get centralized in one place like the US. If most of the world can't afford those chips there will be no returns on the investment until most of the world is globalized again in a somewhat similar living standards (but in a reduced population size).

At least, I don't see the Internet going away. Maybe just in it's current bloated form, because it can still exist once trimmed down and decentralized.

This actually compelled me to make this post on the "Global Depression 2022" thread
I wish to ask a similar question here...
How far do you think the global collapse will set us back in technology?
...and for how long?